infrastructure

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFECREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The Azure Bicep configuration example contains a hardcoded dummy password ('P@ssw0rd123!'). Although explicitly marked as an example for demonstration, it triggers hardcoded credential detection patterns.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) because it provides infrastructure templates that accept external variables.
  • Ingestion points: Variables and parameters in the SKILL.md file, such as 'EnvironmentName', 'appName', and 'app_name'.
  • Boundary markers: The templates do not use explicit delimiters or instructions to prevent an agent from obeying potentially malicious instructions embedded within these variable values.
  • Capability inventory: The skill facilitates the use of highly privileged CLI tools including 'terraform', 'pulumi', 'aws', and 'az', which are capable of modifying and destroying cloud infrastructure.
  • Sanitization: The provided examples do not demonstrate input validation, escaping, or sanitization of the variables before they are used in resource definitions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 06:15 PM
Security Audit — agent-trust-hub — infrastructure